Transcription

Validation and traversal of serialized transcriptions received from the phone. More...

Data Structures
struct	TranscriptionWord
	Word with its confidence value. More...
struct	TranscriptionSentence
	Sentence, a serialized list of words. More...
struct	Transcription
	Transcription: one or more sentences of words. More...

Typedefs
typedef bool(*	TranscriptionSentenceIterateCb) (const TranscriptionSentence *sentence, void *data)
	Sentence iteration callback.
typedef bool(*	TranscriptionWordIterateCb) (const TranscriptionWord *word, void *data)
	Word iteration callback.

Enumerations
enum	TranscriptionType { TranscriptionTypeSentenceList = 0x01 }
	Transcription format. More...

Functions
bool	transcription_validate (const Transcription *transcription, size_t size)
	Check that a transcription received from the phone is well formed.
void *	transcription_iterate_sentences (const TranscriptionSentence *sentences, size_t count, TranscriptionSentenceIterateCb handle_sentence, void *data)
	Iterate over a serialized list of sentences.
void *	transcription_iterate_words (const TranscriptionWord *words, size_t count, TranscriptionWordIterateCb handle_word, void *data)
	Iterate over a serialized list of words.

Detailed Description

Validation and traversal of serialized transcriptions received from the phone.

A transcription is kept in memory exactly as received over the voice endpoint: packed, variable-length sentences, each a list of variable-length words.

static bool count_word(const TranscriptionWord *word, void *data) {
  (*(size_t *)data)++;
  return true;
}
 
if (transcription_validate(t, size)) {
  size_t words = 0;
  const TranscriptionSentence *s = &t->sentences[0];
  transcription_iterate_words(s->words, s->word_count, count_word, &words);
}

---

Data Structure Documentation

TranscriptionWord

struct TranscriptionWord

Word with its confidence value.

Data Fields
uint8_t	confidence	Confidence in percent (1-100), or 0 if not available.
utf8_t	data[]	UTF-8 text, not zero terminated.
uint16_t	length	Length of data in bytes.

TranscriptionSentence

struct TranscriptionSentence

Sentence, a serialized list of words.

Data Fields
uint16_t	word_count	Number of words.
TranscriptionWord	words[]	Serialized words; use transcription_iterate_words() as they have variable length.

Transcription

struct Transcription

Transcription: one or more sentences of words.

Not all recognizers produce several sentences or per-word confidence; the simplest transcription is a single sentence with all confidence values set to 0.

Data Fields
uint8_t	sentence_count	Number of sentences.
TranscriptionSentence	sentences[]	Serialized sentences; use transcription_iterate_sentences() as they have variable length.
TranscriptionType	type: 8	Format of the transcription.

Typedef Documentation

TranscriptionSentenceIterateCb

typedef bool(* TranscriptionSentenceIterateCb) (const TranscriptionSentence *sentence, void *data)

Sentence iteration callback.

Parameters

sentence	Current sentence.
data	Context pointer.

Returns

true to continue, false to stop the iteration.

TranscriptionWordIterateCb

typedef bool(* TranscriptionWordIterateCb) (const TranscriptionWord *word, void *data)

Word iteration callback.

Parameters

word	Current word.
data	Context pointer.

Returns

true to continue, false to stop the iteration.

Enumeration Type Documentation

TranscriptionType

enum TranscriptionType

Transcription format.

Enumerator
TranscriptionTypeSentenceList	List of sentences, the only supported format.

Function Documentation

transcription_iterate_sentences()

void * transcription_iterate_sentences	(	const TranscriptionSentence *	sentences,
		size_t	count,
		TranscriptionSentenceIterateCb	handle_sentence,
		void *	data
	)

Iterate over a serialized list of sentences.

Parameters

sentences	First sentence.
count	Number of sentences.
handle_sentence	Callback for each sentence, may be NULL to only skip over the list.
data	Context pointer passed to handle_sentence.

Returns

End of the list, or the sentence at which the callback stopped the iteration.

transcription_iterate_words()

void * transcription_iterate_words	(	const TranscriptionWord *	words,
		size_t	count,
		TranscriptionWordIterateCb	handle_word,
		void *	data
	)

Iterate over a serialized list of words.

Parameters

words	First word.
count	Number of words.
handle_word	Callback for each word, may be NULL to only skip over the list.
data	Context pointer passed to handle_word.

Returns

End of the list, or the word at which the callback stopped the iteration.

transcription_validate()

bool transcription_validate	(	const Transcription *	transcription,
		size_t	size
	)

Check that a transcription received from the phone is well formed.

The format must be supported, every sentence and word non-empty, words free of control characters other than backspace, and the content must exactly fill size bytes.

Parameters

transcription	Transcription to check, may be NULL.
size	Size of the received buffer in bytes.

Returns

true if the transcription is valid.